the dentine hypersensitivity experience questionnaire: a longitudinal validation study

The Dentine HypersensitivityExperience Questionnaire: alongitudinal validation studyBaker SR, Gibson BJ, Sufi F, Barlow A, Robinson PG. The Dentine HypersensitivityExperience Questionnaire: a longitudinal validation study. J Clin Periodontol 2014;41: 52–59. doi: 10.1111/jcpe.12181.

AbstractAim: To validate the Dentine Hypersensitivity Experience Questionnaire in termsof responsiveness to change and to determine the minimally important difference.Materials and Methods: The study was a secondary analysis of data from threerandomized controlled trials with 311 participants. Three aspects of responsive-ness were examined: change within individuals, differences among people whoimproved, stayed the same or worsened using an external referent and change dueto treatment. Responsiveness to treatments of differing efficacy was assessed intrials with negative and active controls.Results: The measure showed excellent internal reliability, test–retest reliabilityand criterion validity. The measure was highly responsive to change within indi-viduals (Cohen’s effect sizes: 0.28, 0.56, 0.86) showing decreases in the total score(i.e. improvement in OHrQoL) across all trials. The effect sizes in participantswhose self-reported QoL “improved” were large (0.73–1.31). Dentine Hypersensi-tivity Experience Questionnaire detected a treatment effect in one of two negativecontrol trials (effect size: 0.47). Dentine Hypersensitivity Experience Question-naire scores were similar in the test and control groups in the active control trial.The minimally important difference range was between 22 and 39 points.Conclusions: The measure is longitudinally reliable, valid and responsive and candiscriminate between treatments of different efficacy.

Sarah R. Baker1, Barry J. Gibson1,Farzana Sufi2, Ashley Barlow3 and

Peter G. Robinson1

1School of Clinical Dentistry, University of

Sheffield, Sheffield, UK; 2GlaxoSmithKline,

Consumer Healthcare, Weybridge, UK;3GlaxoSmithKline China Investment

Company & TSKF, Shanghai, China

Key words: anti-sensitivity treatment; clinical

trial; dentine hypersensitivity; longitudinal

validation; minimally important difference;

quality of life; responsiveness

Accepted for publication 6 October 2013

Person-reported outcome (PROs)measures are used to capture thepsychosocial experience of pain,discomfort and quality of life, sosupplementing clinical indicators(Jokovic et al. 2002). In oral healthresearch, they identify aspects of lifethat are affected by oral conditions

and that may be amenable to treat-ment (Awad & Feine 1998, Bakeret al. 2006, Pearson et al. 2007).PROs must therefore capturechanges in quality of life. This isreferred to as responsiveness; “theaccurate detection of change when ithas occurred”(DeBruin et al. 1997).

Responsiveness can be examinedin many ways; (1) to detect changewithin individuals, (2) to differentiatebetween subgroups of people whohave improved, stayed the same orworsened on an external anchor orreferent (e.g. clinical endpoint oranother PRO measure) and (3) to

detect treatment or interventioneffects (Revicki et al. 2008). Further-more, to aid interpretation of changethe minimally important difference(MID) is the smallest difference in ascore that a person perceives asimportant (Guyatt et al. 1987). Deter-mining these aspects of responsivenessrequires longitudinal studies thatincorporate interventions of differingefficacy, together with an anchoror criterion (Wyrwich et al. 2005,Revicki et al. 2008).

Dentine hypersensitivity (DH) isa common oral health problem(Cummins 2009) characterized by a

Conflict of interest and source of

funding statement

The authors declare that they have noconflict of interests. The study is sup-ported by a grant from GlaxoSmithK-line Consumer Healthcare R&D.

© 2013 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd52

J Clin Periodontol 2014; 41: 52–59 doi: 10.1111/jcpe.12181

sharp pain in response to an externalstimulus that cannot be explained byany other dental disease (Ayad et al.2009).It is related to the exposure ofthe dentine as a result of gingivalrecession or enamel erosion (Ayadet al. 2009). There has been littleresearch of DH from a patient per-spective. Patients’ experiences ofpain have been recorded in responseto a stimulus within a clinical setting(Al-Wahadni & Linden 2002, Rees& Addy 2002), but there has beenlittle consideration of the impact ofDH on everyday life. Bekes et al.(2009) found that a generic oralhealth-related quality of life (OHr-QoL) measure (OHIP-49) did notdiscriminate the impact of DH;whilst patients with hypersensitivityexperienced more impacts and hadpoorer oral health than the generalpopulation, the difference in meanscores between the two samples wasless than 10% of the overall scale.

A qualitative study explored thedaily experiences of people with DH(Gibson et al. 2010) and showed thecomplexity of pain associated withsensitivity but also impacts on eat-ing, drinking, talking, tooth brushingand social interaction. A condition-specific measure was developed toassess the impact of DH on theOHrQoL (Boiko et al. 2010). ThisDentine Hypersensitivity ExperienceQuestionnaire (DHEQ) detectedfunctional limitations (e.g. slowereating), coping behaviours (e.g.warming food/drinks), emotional(e.g. anxiety) and social impacts (e.g.difficulties conversing) caused byDH and showed excellent reliabilityand validity in both a general popu-lation and a clinical sample (Boikoet al. 2010). It is now necessary toassess the DHEQ’s responsiveness tochange and to interventions.

The aim of this secondary analysiswas to further evaluate DHEQ. Therewere two objectives; (1) to confirm itsreliability and validity and (2) toassess its responsiveness and deter-mine the MID for DHEQ in clinicaltrials. We compared the measures’ability to detect change in two typesof trials; first, two trials comparing atest treatment against negative con-trols, where we hypothesized a priorithat we would detect a treatmenteffect using DHEQ and second, a trialwith an active control which antici-pated no study group difference.

Methods

Participants

In total, 311 people aged 20–65 years took part in three clinicaltrials in the United Kingdom andUnited States (see Table 1).Eachtrial compared the efficacy of testand control dentifrices in providingrelief from DH. Inclusion/exclusioncriteria varied slightly across thethree trials,1 but inclusion criteriacommonly included at least twonon-adjacent sensitive teeth with aSchiff Sensitivity Score of ≥2 and atactile threshold (Yeaple probe)of ≤20 g force. Exclusion criteriaincluded (1) teeth with evidence ofcurrent or recent caries, or reportedtreatment of decay in 12 months ofscreening and (2) teeth with exposeddentine but with deep, defective orfacial restorations, teeth used asabutments for fixed or removablepartial dentures, teeth with fullcrowns or veneers, orthodonticbands or cracked enamel. Sensitiveteeth with contributing aetiologiesother than erosion, abrasion orrecession of exposed dentine.

The studies were approved by anInstitutional Review Board or Inde-pendent Ethics Committee in accor-dance with local requirements.

Clinical trial overview

All three trials were randomized,examiner-blinded, two treatment arm,stratified, parallel design and single-site studies. Demographic, medicalhistory and medications were recordedfollowed by oral examination. Partici-pants returned at least 24 h afterscreening for baseline assessment,where they completed DHEQ, andstandard tooth sensitivity assessment.

Participants were then randomized(1:1 allocation) according to baselinestratification, dispensed their studydentifrice, standard toothbrush,diary, brushing instructions and tim-ers. Participants brushed at hometwice daily for 8 (Trials B and C) or12 weeks (Trial A). Participantsreturned for Week 4, 8 (Trials B andC) and 12 visits (Trial A).

The revised DHEQ (Boiko et al.2010) has 48 items of which 34 com-prise an impact scale (see Table 2).The scale contains five domains ofFunctional restrictions (4 items), Cop-ing (12), Social impact (5), Emotionalimpact (8) and Identity (5). The itemshave coded responses on 7-pointLikert scales: 1 = “strongly disagree”,2 = “disagree”, 3 = “agree a little”,4 = “neither agree nor disagree”, 5= “disagree a little”, 6 = “disagree”and 7 = “strongly disagree”. A sum-mary measure, “total score” wascalculated as the sum of the scoresper participant, with a possible rangeof 34–238. Domain scores were calcu-lated in the same way.

Table 1. Participants’ demographics across the three clinical trials at baseline

Trial A Trial B Trial C

N 93 118 100Control 46 58 49Test 47 60 51GenderFemale 75 83 81Male 18 35 19

AgeMean 42.90 36.19 45.21SD, Range 10.18, 20–60 10.71, 21–65 9.50, 20–65

Duration of DH< 6 months 1 1 7>6 < 12 months 8 14 6>1 < 5 years 36 62 53>5 < 20 years 33 41 29>20 years 15 0 4

DH intensity*: Mean (SD) 6.25 (1.57) 6.46 (1.29) 5.80 (1.76)Tactile: Mean (SD) 12.84 (3.46) 10.00 (0.00) 12.35 (3.05)VAS: Mean (SD) 69.45 (13.07) – 57.26 (22.07)Schiff: Mean (SD) 2.76 (0.338) 2.71 (0.366) 2.69 (0.401)

*Potential range 1(not at all) – 10(extremely).

1Available on request.

© 2013 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd

Dentine Hypersensitivity Experience Questionnaire 53

Participants also self-report theirglobal oral health on a 6-point scalefrom 1 (excellent) to 6 (very poor).The effect of DH on life overall wasmeasured using four items ratedfrom 0 (not at all) to 4 (very much)creating a possible range 0–16.

The Evaporative (air) sensitivitywith Schiff sensitivity score is madeby directing a 1 sec application ofair from a dental syringe to thetooth surface 1–2 mm coronal to thefree gingival margin from 1 cm. Theresponse was rated from 0–3 usingthe Schiff sensitivity scale (Schiffet al. 1994).

A Visual analogue scale (VAS):rated participants response to theevaporative air test on a 100 mmscale scored from 0 (no pain) to 100(intense pain).

The Tactile sensitivity (Yeapleprobe, Polson et al. 1980) score wasmade by directing the probe tip per-

pendicular to the buccal surface andslowly drawing it across the toothsurface with a force setting of 10 g.This is increased by 10 g, with eachsuccessive challenge, until either a“yes” response is recorded or themaximum reached (20 g – Trials Band C; 30 g Trial A – baselinesettings only). The gram settingthat elicited two consecutive “yes”responses was recorded as the tactilethreshold.

Data analytic strategy

The analysis was conducted in twostages. In the first stage, cross-sec-tional validation was carried out tofurther examine the reliability andvalidity of the DHEQ. Firstly, itemimpact was calculated as mean scoremultiplied by the percentage of peo-ple who reported an impact (a“agree a little”,“agree” or “strongly

agree” response on the item) to high-light the DH impacts that were mostimportant for most people. Sec-ondly, internal consistency of theDHEQ was assessed via item-totalcorrelations, subscale-total correla-tions and Cronbach’s alpha coeffi-cient. Thirdly, test–retest reliabilitywas assessed with intra-class coeffi-cients (ICC) using data from 41control group participants who pro-vided follow-up data at 12 weeks.Lastly, convergent validity was exam-ined by correlating DHEQ scoreswith global oral health ratings, self-reported quality of life and clinicalmeasures.

In the second stage of the analy-sis, the longitudinal validation wascarried out. Firstly, we examinedresponsiveness over time within indi-viduals by calculating the effect sizefor individual changes between base-line and follow-up as follows:

Table 2. Mean (SD) scores, item impacts and item-total correlations of DHEQ items at baseline for clinical trial A

Mean SD Item impact Item total correl. (r)

1 Restrictions - pleasure out of eating 5.14 1.56 397.32* 0.7372 Restrictions - cannot finish meal 3.92 1.62 168.95 0.7593 Restrictions - longer to finish meal 5.06 1.61 375.45 0.7524 Restrictions - problems eating ice-cream 6.11 0.97 584.73* 0.5725 Approach coping - modification in eating 5.55 1.36 489.51* 0.6796 Approach coping - careful when breathing 5.61 1.16 494.80* 0.6247 Approach coping - warming food/drinks 5.04 1.69 368.42 0.6878 Approach coping - cooling food/drink 4.43 1.79 233.46 0.6319 Approach coping - cutting fruit 4.72 1.78 289.34 0.61610 Approach coping - putting a scarf over mouth 4.61 1.75 258.16 0.44411 Avoidant coping - cold drinks/foods 5.08 1.56 430.36* 0.67412 Avoidant coping - hot drinks/foods 4.06 1.73 183.11 0.53213 Avoidant coping - contact with certain teeth 5.47 1.37 453.46* 0.60314 Avoidant coping - change toothbrushing 4.46 1.53 249.31 0.45915 Avoidant coping - biting in small pieces 5.15 1.51 387.80 0.61016 Avoidant coping - other food 4.52 1.76 262.16 0.82717 Social - longer than others to finish 4.29 1.78 217.07* 0.70418 Social - choose food with others 3.99 1.72 158.80 0.83719 Social -hide the way of eating 3.66 1.64 121.88 0.79420 Social - unable to take part in conversations 2.73 1.50 32.49 0.62321 Social - painful at the dentist 4.96 1.78 352.16* 0.48522 Emotions - frustrated not finding a cure 4.77 1.55 297.65 0.66923 Emotions - anxious of eating contributes 5.10 1.50 384.03 0.72324 Emotions - irritating sensations 5.83 1.05 564.34* 0.75225 Emotions - annoyed I contributed 4.32 1.78 204.34 0.50926 Emotions - guilty for contributing 3.82 1.69 144.01 0.55027 Emotions - annoying sensations 5.76 1.02 551.23* 0.72128 Emotions - embarrassing sensations 3.43 1.76 99.81 0.77629 Emotions - anxious because of sensation 4.41 1.62 256.22 0.71230 Identity - difficult to accept 3.28 1.74 95.45 0.60231 Identity - different from others 3.26 1.71 101.39 0.71632 Identity - makes me feel old 3.81 1.88 164.21* 0.47833 Identity - makes me feel damaged 3.30 1.73 99.33 0.63334 Identity - makes me feel unhealthy 3.65 1.79 164.62* 0.614

*Two items with the greatest impact within each domain (restrictions, coping behaviours, social impacts, emotional impacts and identity).Figures in bold are those items with the greatest impact across the whole scale.


54 Baker et al.

Within-person individual effectsize =

Post-test DHEQ score

� Pre-test DHEQ score

Pre-test (group) standard deviation

Within-person individual effect sizesfrom 0.2 to 0.49 were classified assmall, from 0.5 to 0.79 as moderateand greater than 0.8 as large (Cohen1998).

Secondly, we assessed responsive-ness using an external referent inthree ways in line with the literature(Revicki et al. 2008). First, partici-pants were classified by their changein self-reported QoL from baselineto final follow-up (i.e. better, sameor worse). Second, we assessedresponsiveness to change indices;namely, Cohen’s effect size (ES) andthe standardized response mean(SRM) (Guyatt et al. 1987). ES wascalculated as the mean changebetween baseline and final follow-updivided by the baseline standarddeviation (SD). SRM was calculatedas the mean change between baselineand follow-up divided by the SD ofthe change score. Both the ES andSRM were calculated stratified bycategory of change (worse, same orbetter). Lastly, to aid interpretationof DHEQ change scores for futureclinical trials, we calculated theMID. The MID was defined as themean change in the total scores inparticipants who reported anyimprovement in their self-reportedQoL (i.e. impact on life overall).

Thirdly, we calculated responsive-ness to treatment. If the DHEQ isresponsive to treatments for DH, itshould be able to detect an effect oftreatments with differing efficacies.Responsiveness was comparedbetween trials with negative controls(Trials A and B where our a priorihypothesis was that DHEQ changescores would differ between test andcontrol groups) and an active con-trol trial (C where it was anticipatedto detect similar changes in bothgroups).

Three analyses were conducted,as recommended by Buchbinderet al. (1995); (1) the observed treat-ment effect was calculated as the dif-ference between mean change scoresin the test and control groups,

(2) the relative per cent improve-ment as

Test change � control change

Mean test baseline

� mean control baseline

� 100

and (3) the standardized effect sizesby trial arm were calculated as theobserved treatment effect divided bythe SD of the pooled change scores.

Results

Cross-sectional validation

The greatest impacts of DH on dailylife were problems with eating ice-cream, irritating sensations, annoyingsensations, modification in eating,care when breathing, avoidance ofcold drinks or foods and avoidingcontact with certain teeth (seeTable 2). With regard to internal con-sistency, nearly all item-total correla-tions were over 0.4 indicating goodinternal consistency (see Table 2). Allcorrelations between domains andtotal scores were significant (atp < 0.05) and consistent. The correla-tions between total score and thedomains were as follows: socialimpact (r = 0.905), coping (0.911),restrictions (0.876), emotional impact(0.898) and identity (0.756).Internalconsistency for the total score(a = 0.96) and the domains of restric-tions (0.85), coping (0.91), socialimpact (0.84), emotional impact (0.89)and identity (0.90) were all excellent.

Test–retest reliability was high forthe total impact score at 0.77, butvaried for the domains: identity

(0.89), social impact (0.72), emo-tional impact (0.68), coping (0.61)and restrictions (0.43). All were sig-nificant at p < 0.001, except for therestrictions domain at p < 0.01. Withregard to convergent validity, DHEQtotal score was not correlated withglobal oral health ratings, or therestrictions, coping or emotionsdomains (see Table 3). There weresignificant correlations between theidentity and social impact domainsand the global oral health rating.DHEQ total and all domain scoreswere strongly and significantly corre-lated with the scores for effect ofDH on life overall. The total DHEQand domain scores were significantlycorrelated with the mean tactilescores but not to the VAS and Schiffsensitivity assessment.

Longitudinal validation

Responsiveness over time within indi-viduals

DHEQ total and domain scores inall three trials are summarized inTable 4. There were significantdecreases in total scores and mostdomain scores across all trials. Theidentity domain consistently showedthe lowest change followed by thesocial domain. Within-person indi-vidual effects for total DHEQ scoreswere large in the active control trial(C) (0.86) and small to moderate inthe negative control trials (0.28, 0.56A and B respectively). The copingdomain showed the largest effect sizein all trials (0.31, 0.63, 0.90 A, Band C respectively) followed by theemotional domain (0.23, 0.64, 0.75).The identity domain showed thesmallest (0.18, 0.10, 0.27).

Table 3. Pearson’s correlation coefficients between DHEQ total and domain scores andself-reported global oral health, quality of life and clinical measures at baseline in clinicaltrial A

Self-reported Clinical

Global oral health QoL Tactile VAS Schiff†

DHEQ total 0.190 0.793*** �0.34*** 0.16 0.20Restrictions 0.149 0.677*** �0.32** 0.11 0.10Coping 0.042 0.700*** �0.30** 0.20 0.28**Social 0.292** 0.720*** �0.27** 0.10 0.12Emotional 0.166 0.718*** �0.34*** 0.13 0.11Identity 0.287** 0.640*** �0.24* 0.09 0.17

*p < 0.05, **p ≤ 0.01, ***p ≤ 0.001.†Spearman’s correlation coefficient.



Responsiveness using an external ref-erent

A total of 58, 78 and 84 participantsreported improved DHEQ totalscores, with 22, 30 and 12 reportingdeterioration in Trials A, B and Crespectively. Only nine participantsreported no change from baseline tofollow-up across the three trials.

To examine responsiveness usingan external criterion, participantswere classified by their change inself-reported QoL from baseline tofinal follow-up (i.e. better, same orworse). One-way analysis of variancewas used to compare change inDHEQ scores across these threegroups. There were main effects ofchange group in all three trials(F = 12.34, 8.24, 14.02 all p-val-ues<0.001). If the DHEQ is respon-sive to change, it should showchanges in participants who reportedimprovement in QoL and theseshould be substantially greater thanthose demonstrating stability in theexternal referent. The results werecompatible with these expectations;those participants who reported animprovement in QoL at follow-up(the “better” group) had more

improvement in DHEQ scores (i.e.less DHEQ impacts over time) thanthose in the “same” and “worse”groups (see Table 5). In the activecontrol trial (C) all groups showedpositive change scores (improvementin OHrQoL) regardless of whethercategorized as “better”, “same” or“worse”. These findings are compati-ble with expectations; as all partici-pants received active treatment.

Using responsiveness to changeindices, we found that in the nega-tive control trials, both the ES andSRM could be classified as largeeffect sizes in the “better” group(>0.80, Cohen 1998) with very smalleffect sizes in both “same” and“worse” groups. In the active controltrial, all effect sizes were large (seeTable 5).

Finally, the MID was calculatedas the mean change in the total scoresin participants who reported anyimprovement in their self-reportedQoL (i.e. impact on life overall). Forthe negative control trials, the MIDrange was between 22 (A, n = 41) and29 points (B, n = 71); whilst for theactive control trial, the MID was lar-ger at 39 points (C, n = 55).

Responsiveness to treatment

DHEQ detected a treatment effect inthe first negative control trial (A).Participants who received the activetreatment reported greater reductionin DHEQ scores at follow-up thancontrols (Table 6). The interventioneffect was moderate (Cohen’s classi-fication 0.47). Contrary to ourhypothesis, changes in both groupswere similar in the second nega-tive control trial (B). The relativeimprovement in this trial was 2.1%,with a very small effect size (0.12).As hypothesized, changes in DHEQscores were similar in the test andcontrol groups in the active controltrial. The relative improvement was5.1% in the control compared withtest group, with a small effect size(�0.25).

Table 4. Participants DHEQ total, domain(mean, SD) and change scores across clinical tri-als

Baseline 8-week 12-week Pre-post EffectChange† size

Trial ATotal 151.54 (36.05) – 141.41 (38.34) 10.13*** 0.28Restrictions 20.01 (4.84) – 19.13 (5.22) 0.88 0.18Coping 58.26 (13.53) – 54.12 (15.98) 4.14*** 0.31Social 19.42 (6.46) – 17.71 (6.60) 1.70*** 0.26Emotional 36.82 (9.16) – 34.73 (9.00) 2.10* 0.23Identity 17.02 (7.45) – 15.71 (7.24) 1.31** 0.18

Trial BTotal 150.85 (36.79) 130.48 (43.49) – 20.38*** 0.56Restrictions 20.10 (5.07) 17.18 (5.59) – 2.92*** 0.58Coping 57.01 (13.15) 48.77 (15.97) – 8.24*** 0.63Social 19.66 (6.39) 17.22 (6.99) – 2.44*** 0.38motional 37.55 (9.92) 31.30 (10.62) – 6.25*** 0.64Identity 16.80 (8.60) 15.91 (7.99) – 0.90 0.10

Trial CTotal 136.69 (30.96) 108.84 (39.17) – 27.84*** 0.86Restrictions 19.15 (4.49) 14.66 (5.52) – 4.49*** 0.99Coping 52.64 (11.98) 41.79 (15.67) – 10.86*** 0.90Social 16.65 (6.29) 13.21 (5.98) – 3.43*** 0.54Emotional 34.90 (9.25) 27.87 (11.49) – 7.03*** 0.75Identity 14.57 (6.88) 12.69 (6.67) – 1.88*** 0.27

*p < 0.05, **p ≤ 0.01, ***p ≤ 0.001.†Paired t-test.Higher score, worse OHrQoL; Trial A and B, negative control; Trial C, active control.

Table 5. DHEQ total change scores, effect size and standardized response means accordingto QoL outcome category at follow-up

n Change (SD) Standardized Effectresponse mean size

Trial ABetter 41 22.39 (22.58) 0.99 0.73Same 17 2.53 (23.22) 0.11 0.05Worse 26 -4.23 (21.92) 0.19 0.12

Trial BBetter 71 29.44 (34.73) 0.85 0.84Same 13 3.31 (21.39) 0.16 0.09Worse 25 3.52 (27.09) 0.13 0.08

Trial CBetter 55 39.38 (28.67) 1.37 1.31Same 25 16.64 (14.37) 1.16 0.49Worse 15 28.54 (7.37) 3.87 0.81

DHEQ scores recoded so higher score, improvement in OHrQoL; Trial A and B, negativecontrol; Trial C, active control.


56 Baker et al.

Discussion

The application of condition-specificperson-reported measures, such asthe DHEQ, as outcome measures inrandomized clinical trials is relativelynew in dentistry. As such, it is par-ticularly important that any newlydeveloped – or indeed existing –PRO measure meets the gold stan-dard guidelines for establishingresponsiveness or sensitivity tochange; a key component of anyevaluative measure (Guyatt et al.1987). Using the methods recom-mended in recent guidelines (Revickiet al. 2008) and by the clinical signif-icance consensus group (Wyrwichet al. 2005), we demonstrated thatthe DHEQ is highly responsive tochanges in functional and personalexperiences of DH within individualsover an 8 to 12 week timeframe.There was significant improvementin DHEQ scores in all three trials.These findings indicate that theDHEQ is valuable for use in longitu-dinal evaluation of people with DH.Interestingly, the coping domainshowed the greatest change - com-patible with these items being tran-sient or amenable to change (e.g.warming certain foods or drinks;avoiding cold drinks). In contrast,the identity and social domains wereleast responsive perhaps becausethey are less likely to change overthe short-term (e.g. makes me feelold; damaged).This differential pat-tern of change was reflected in theeffect sizes for the individual domainscores. Overall, for the total DHEQscore, the effect sizes were small tomoderate in the negative control tri-als (0.26–0.56) and large in the activecontrol trial (0.86). This findingwould be expected; if both con-trol and test groups received anti-sensitivity dentifrice they would beexpected to show the greatestimprovement in OHrQoL (Table 4).

Using changes in self-reportedQoL as the external referent, theDHEQ was sensitive to improvementin QoL status. Indeed, ES andSRM (ranges: 0.73–1.31) for the“improved” group represent largeeffects according to Cohen’s classifi-cation, which suggests robust respon-siveness. Similarly, the small SRMand ES in the stable group inthe negative control trials (ranges:0.05–0.09) provide further supportfor the stability of the DHEQ.

When physiological or clinicalmeasures are tested there is often agold standard for comparison. Inmost cases relating to PRO measuressuch as QoL there is, however, nogold standard. Given that we do nothave a gold standard in relation toDH, we could not examine clinicallysignificant change in DHEQ; that is,“a difference score that is largeenough to have an implication forthe patient’s treatment or care”(Wyrwich et al. 2005). This hasimportant implications for determin-ing the MID; a statistic or cut-pointwhich is vital for interpretingOHrQoL scores in both cross-sec-tional comparisons of groups andlongitudinal evaluation of treatmentstrategies. Here, we chose to use par-ticipants’ self-reported overall QoL.In accordance with the clinical sig-nificance consensus group, we wouldargue that the person themselvesshould be the ultimate judge of theirhealth status. Thus, a person’s self-report should be a better anchor forestimating MIDs of QoL measures(Wyrwich et al. 2005).In this study,the MID was estimated at 22, 29and 39 in the three trials. Theseresults mean that if the measure wereto be used in further clinical trials aDHEQ total change score between22 and 29 would be required in neg-ative control trials, with a largerchange score of 39 in active controltrials. As with all PRO measures, the

MID is not an immutable character-istic; it will vary across populationsand treatments (Revicki et al. 2008).Nevertheless, the MID presentedhere is important for the interpreta-tion of statistically significant resultsin future DH-related clinical trials,as well as determining future samplesizes for clinical trials using DHEQ.

The findings suggest that DHEQis responsive to treatments, support-ing our a priori hypotheses in two ofthe three trials. In the first,participants using an active dentineanti-sensitivity dentifrice reportedsignificantly fewer impacts of DH ontheir daily lives at 12 week follow-upcompared with those who receivedno anti-sensitivity ingredient(Table 6). The responsiveness totreatment was supported by theactive control trial; improvements intest and control groups were similar,again supporting our hypothesis thatparticipants should improve regard-less of group. In contrast, DHEQdid not detect an effect in the secondnegative control trial (B). The expla-nation of this remains unclear as theinclusion criteria and study designwere comparable in both negativecontrol trials. However, controls intrial B show much greater changescores than in trial A (18.21 versus4.05 respectively) and the standarddeviations were much greater in testand control groups (34.16, 33.86respectively) compared with Trial A(26.09, 23.45). These changes in thecontrol group influence responsive-ness statistics. Furthermore, it isimportant to note that the trialswere powered to detect changes inSchiff scores, as the primary out-come, rather than DHEQ.

The clinical significance consensusgroup commented that evaluations ofPROs, particularly QoL in clinicaltrials and the assessment of clinicallysignificant thresholds for change arecomplex (Wyrwich et al. 2005).Indeed, strategies to determinechange have not kept pace with theexplosion in QoL measures in medi-cine and dentistry. Whilst we usedseveral analytic strategies, DHEQ,like all such measures should be con-stantly reviewed, incorporating newdata and psychometric techniques asthey emerge. We used both anchor-based (i.e. tied to an external refer-ent) and distribution-based methods(i.e. linked to a statistical parameter

Table 6. DHEQ total change scores, treatment effect and standardized effect size oftreatment among trial participants

Test (SD) Control (SD) F Observed Relative % SESchange change treatment improvement

effect

Trial A 15.93 (26.09) 4.05 (23.45) 4.80* 11.88 7.68 0.47Trial B 22.43 (34.16) 18.21 (33.86) 0.42 4.22 2.05 0.12Trial C 24.35 (24.98) 31.49 (32.35) 1.47 �7.14 �5.14 �0.25

*p < 0.05.Trial A and B = negative control; Trial C = active control.



such as SD) as outlined at the con-sensus meeting (see Table 3, Wyr-wich et al. 2005). Despite theirwidespread use, responsiveness statis-tics are derived using SDs from agiven sample and are therefore popu-lation and context specific. We miti-gated this to some degree bycalculating responsiveness acrossthree trials incorporating both activeand negative control arms. Neverthe-less, the MID identified here may notbe appropriate outside of clinical tri-als, such as people with mild DH inthe general population. In addition,we calculated the MID based onactual changes in QoL rather thanglobal ratings of change in health.The latter, most commonly usedmethod for determining MID, wasnot available in the data set. Giventhat we have no prior information onwhat constitutes a minimal change inthe anchor (the effect of DH on lifeoverall), we selected a threshold ofany change. Whilst this strategy isacceptable, it does risk over-estimat-ing the MID (Revicki et al. 2008).

The DHEQ was developed bymeticulously following recommendedmethods and adopting a robust theo-retical framework (Wilson & Cleary1995, Juniper et al. 1996).This studysupports findings from earlier cross-sectional validation (Boiko et al.2010) and a study in China (Heet al. 2012).This body of work dem-onstrates the DHEQ to be a reliableand valid measure of the experienceof DH and a useful addition to thegrowing condition-specific OHrQoLinstruments for facilitating ourunderstanding of the biopsychosocialimpacts of oral conditions. DHEQ istherefore the measure of choicebecause of its direct reference to theproblems associated with sensitiveteeth. It can therefore help toexplore the nuances of impacts inrelation to dentine hypersensitivityspecifically which, in turn, has possi-ble implications for our futureunderstanding of a condition that issometimes considered an enigma(Johnson et al. 1982) in addition to

serving as a tool for use in clinicaltrials.

References

Al-Wahadni, A. & Linden, G. J. (2002) Dentinehypersensitivity in Jordanian dental attenders -A case control study. Journal of Clinical Peri-odontology 29, 688–693.

Awad, M. A. & Feine, J. S. (1998) Measuringpatient satisfaction with mandibular prostheses.Community Dentistry and Oral Epidemiology26, 400–405.

Ayad, F., Ayad, D., Zhang, Y. P., DeVizio, W.,Cummins, D. & Mateo, L. R. (2009) compar-ing the efficacy in redcuing dentin hypersensi-tivity of a new toothpaste containing 8/0%arginine, calcium carbonate, and 1450 ppmflouride to a commercial sensitive toothpastecontaining 2% potassium ion: an eight-weekclinical study on Canadian adults. The Journalof clinical dentistry 20, 10–16.

Baker, S. R., Pankhurst, C. L. & Robinson, P. G.(2006) Utility of two oral health-related qualityof life measures in patients with xerostomia.Community Dentistry and Oral Epidemiology34, 351–362.

Bekes, K., Johh, M. T., Schaller, H.-G. & Hirsch,C. (2009) Oral health-related quality of life inpatients seeking care for dentin hypersensitiv-ity. Journal of Oral Rehabilitation 36, 45–51.

Boiko, O. V., Baker, S. R., Gibson, B. J., Locker,D., Sufi, F., Barlow, A. P. S. & Robinson, P.G. (2010) Construction and Validation of theQuality of Life Measure for Dentine Hypersen-sitivity (DHEQ). Journal of Clinical Periodon-tology 37, 973–980.

Buchbinder, R., Bombardier, C., Yeung, M. &Tugwell, P. (1995) Which outcome measureshould be used in rheumatoid arthritis clinicaltrials? Arthritis and Rheumatism-Arthritis Careand Research 38, 1221–1222.

Cohen, J. (1998) Statistical power analysis for thebehavioural sciences. Second Edition.NJ: Law-rence Erlbaum Associates.

Cummins, D. (2009) Dentin Hypersensitivity:From Diagnosis to a Breakthrough Therapyfor Everyday Sensitivity Relief. The Journal ofclinical dentistry 20, 1–9.

DeBruin, A. F., Diederiks, J. P. M., De Witte, L.P., Stevens, F. C. J. & Philipsen, H. (1997)Assessing the responsiveness of a functionalstatus measure: the sickness impact profile ver-sus the SIP68. Journal of Clinical Epidemiology50, 529–540.

Gibson, B., Boiko, O. V., Baker, S. R., Robinson,P. G., Barlow, A., Player, T. & Locker, D.(2010) The everyday impact of dentine sensitiv-ity: personal and functional aspects. Social Sci-ence and Dentistry 1, 11–21.

Guyatt, G. H., Walter, S. & Norman, G. (1987)Measuring change over time: assessing the use-fulness of evaluative instruments. Journal ofChronic Disease 40, 171–178.

He, S. L., Wang, J. H. & Wang, M. H. (2012)Development of the Chinese version of the

Dentine Hypersensitivity Experience Question-naire. European Journal of Oral Sciences 120,218–223.

Johnson, R. H., Zulgar-Nairn, B. J. & Koval, J.J. (1982) The effectiveness of an electroniosingtoothbrush in the control of dentinal hypersen-sitivity. Journal of Periodontology 53, 353–359.

Jokovic, A., Locker, D., Stephens, M., Kenny,D., Tompson, B. & Guyatt, G. (2002) Validityand reliability of a questionnaire for measuringchild oral-health-related quality of life. Journalof Dental Research 81, 459–463.

Juniper, E. F., Guyatt, G. H. & Jaeschke, R.(1996) How to Develop and Validate a NewHealth-Related Quality of Life Instrument.Quality of Life and Pharmacoeconomics in Clin-ical Trials. Philadelphia: Lippincott-Raven.

Pearson, N. K., Gibson, B. J., Davis, D. M., Gei-lier, S. & Robinson, P. G. (2007) The effect ofa domiciliary denture service on oral healthrelated quality of life: a randomised controlledtrial. British Dental Journal 203, E3, 568.

Polson, A. M., Caton, J. G., Yeaple, R. N. &Zander, H. A. (1980) Histological determina-tion of probe tip penetration into gingivalsulcus of humans using an electronic pressure-sensitive probe. Journal of Clinical Periodontol-ogy 7, 479–488.

Rees, J. S. & Addy, M. (2002) A cross-sectionalstudy of dentine hypersensitivity. Journal ofClinical Periodontology 29, 997–1003.

Revicki, D., Hays, R. D., Cella, D. & Sloan, J.(2008) Recommended methods for determiningresponsiveness and minimally important differ-ences for patient-reported outcomes. Journal ofClinical Epidemiology 61, 102–109.

Schiff, T., Dotson, M., Cohen, S., De Vizio, W.,McCool, J. & Volpe, A. (1994) Efficacy of adentifrice containing potassium nitrate, solublepyrophosphate, PVM/MA copolymer, andsodium fluoride on dentinal hypersensitivity: atwelve week clinical study. The Journal of clini-cal dentistry 5, 87–92.

Wilson, I. & Cleary, P. (1995) Linking clinicalvariables with health-related quality of life: aconceptual model of patient outcomes. Journalof the American Medical Association 273, 59–65.

Wyrwich, K. W, Bullinger, M, Aaronson, N,Hays, R. D, Patrick, D. L & Symonds, T. andthe clinical significance consensus meetinggroup. (2005) Estimating clinically significantdifferences in quality of life outcomes. QualityLife Research 14, 285–295.

Address:Sarah BakerUnit of Dental Public HealthSchool of Clinical DentistryUniversity of SheffieldS10 2TAUKE-mail: [email protected]


58 Baker et al.

Clinical Relevance

Scientific rationale for the study:The application of condition-spe-cific person-reported measures asoutcomes in clinical trials is rela-tively new in dentistry. It is impor-tant that any newly developedevaluative measure meets the goldstandard guidelines for establishingresponsiveness. This study soughtto assess the responsiveness of the

Dentine Hypersensitivity ExperienceQuestionnaire (DHEQ) in clinicaltrials.Principal findings: The results dem-onstrate the DHEQ to be highlyresponsive to changes in functionaland personal experiences of DH,improvement in quality of life statusand to anti-sensitivity treatmentswith differing efficacies.

Practical implications: The studyprovides a comprehensive psycho-metric evaluation of the DHEQ. Itoffers clinicians, trial lists andresearchers a tool to understandand quantify the experience of DHin everyday life and the effects oftreatment from the person perspec-tive.



the dentine hypersensitivity experience questionnaire: a longitudinal validation study

Documents